Pronouns/ergative languages

Alexander S. Nikolaev alex at AN3039.spb.edu
Thu Jun 10 17:35:42 UTC 1999


Nicholas Widdows wrote:

<I don't count direct object and intransitive subject being marked
<the same, even if it is in inanimate nouns and therefore at the end of the
<Silverstein hierarchy most likely to be ergative.

I should like to note, that since in all stems except for thematic
ones neuters have no marker at all, it wouldn't seem very plausible
to regard them as ergative-marked.

Anyway, as far as Silverstein's hierarchy is concerned, i'd like to
put here the following idea:
As there really are some grounds to reconstruct genitivus-ergativus,
which was already mentioned in the discussion, one could throw a
glance at the typological data; the issue will be, that languages,
which possess a case, *combining* the two functions, namely ergative
and oblique, differ substantially from the languages, which are
characterized by independent ergative case. The former are, to say
roughly, less "ergative" and are very often diachronically on the way
of "accusativization" (or, vice versa, fuller "ergativization").
They take an intermediate position in the
continuum "ergative -- accusative"; and very often they are
characterized by the tense-split, and the ergative system in such
languages is confined to preterit tense(s). Such are Eskimo, Lach,
Burushaski, Kurdish (which, i know, is not a very good example, since
it developed a secondary ergative-like structure, but from a synchronic
point of view it will do). [There is, besides, a scanty piece of
evidence for tense-split in PIE].
	And in such semi-ergative or semi-accusative languages
Silverstein's hierarchy just doesn't work, being
a property of "full-ergative" languages (I apologize for my "terms",
which are not terminological at all, and, of course, i bear all the
responsibility for them).
		Thus one can assume that it didn't
work in PIE either, and so the reasons which led Villar and Rumsey to
reject ergativity for PIE on the basis of this universal are not
important anymore, and hence one is entitled to believe that
inanimates just took the ergative marker less often, than animates. I
foresee the possible objection "but *all* the nouns with no respect
to gender are marked with genitive marker *os/*es/*s!" This i'm
inclined to explain as a consequence of tense-split structure of PIE:
this case in *os/*es/*s was common for all the nouns in present tense
and had the meaning of genitive; and the same marker performed the
functions of ergative case in preterit, and only animates could take
it, hence the examples of genitive of agent in historical IE dialects.
	This intermediate stage of PIE, which, i believe, can be
reconstrructed within the framework of internal reconstruction method
and is *not* based on pure speculations, could of course be preceded
by another stage, when PIE was characterized by other structure, e.g.
active. The residues of the latter can be seen in the two-series
verbal system.

[To drop a couple of names more, among those
who insisted on genitive case as reflecting the proto-agent,
are Holger Pedersen, Jerzy Kurylowicz, Robert Beekes,
William Schmalstieg, Silvia Luraghi et al.]

I think, this idea could account for many oddities of PIE nominal system,
though i have to agree, that we have no evidence of Agens being
marked otherwise than Subject, i.e. the 1st actant of 1-argument
predicate.

I would be thankful for any kind of remarks and corrections.

Regards,
Alex



More information about the Indo-european mailing list